Cairo Henrique Bueno

Anthropic’s Annual Recurring Revenue (ARR) has reportedly reached an astounding $44 billion, representing a nearly 5x (389%) increase from the $9 billion recorded at the close of 2025.

Bach-1.0 Preview from Video Rebirth debuts at #6 on the Artificial Analysis Text to Video Leaderboard (No Audio)! Bach-1.0 Preview is the latest Text to Video model from @video_rebirth, with similar performance to Vidu Q3 Pro, Kling 3.0 Omni 1080p (Pro), and grok-imagine-video. Bach-1.0 Preview is intended for broad release later in May. See example generations from Bach-1.0 Preview in the Artificial Analysis Video Arena below 🧵

OpenAI’s GPT-5.5 has become the second model to successfully complete a complex, 32-step end-to-end corporate network attack simulation. The model demonstrated an average 71% success rate on expert-level cybersecurity challenges, including memory corruption exploitation, breaking cryptographic implementations, and reversing stripped binaries. In a benchmark involving the reverse-engineering of a custom virtual machine a task that took a human expert roughly 12 hours, GPT-5.5 solved the challenge in under 11 minutes at a total compute cost of just $1.73. In a full-scale simulation representing approximately 20 hours of human expert effort, GPT-5.5 achieved a total network takeover in 2 out of 10 attempts.

xAI released voice cloning capabilities on the xAI Console for users in the United States. Users can now create a custom voice in under two minutes. Additionally, the platform offers a library of over 80 pre-set voices supporting 28 different languages.

We implemented @karpathy 's MicroGPT fully on FPGA fabric. No GPU. No PyTorch. No CPU inference loop. Just a transformer burned into hardware, generating 50,000+ tokens/sec. The model is small, but the idea is not: inference does not have to live only in software 👇

THIS GUY BUILT A TOOL THAT LETS YOU CREATE INTERACTIVE 3D TOURS WITH JUST AN IPHONE you walk around a room with your iphone, the app scans it, and it spits out a fully interactive 3D tour in minutes no matterport camera, no $3000 equipment, and no photographer needed here's how people are already planning to use it: a german shipping container home builder wants to use it to scan empty containers and visualize custom interiors for clients before they build a homeowner is using it to scan his 100+ year old unfinished basement to plan a renovation and see what the finished space would look like another guy walked around the outside of a trailer to get a full external 3D model real estate agents, interior designers, renovation companies, airbnb hosts, property managers, and architects can all use this every single one of these people currently pays $200-500 per property for matterport scans or hires a photographer to do it this replaces that entire workflow with an iphone and a few minutes of walking around

Wow. This is huge "Researchers developed a DNA-based therapy using polypurine reverse Hoogsteen hairpins (PPRHs) to directly silence the PCSK9 gene, a key regulator of LDL cholesterol." "In human liver cells, the therapy reduced PCSK9 protein levels by up to 87%, showing strong gene-silencing efficiency." "In transgenic mice expressing human PCSK9, a single dose lowered total cholesterol by 47% within 3 days." "Unlike statins or antibody drugs, this approach works at the DNA level, blocking gene transcription before the protein is even produced." "The treatment showed no significant toxicity in animal models and maintained stability with low immunogenic response."

Last week, we made Gemini Embedding 2, our first natively multimodal embedding model, available to the general public. Since then, developers have used it to build video analysis tools, visual shopping assistants, and more. But you might be wondering... what is an embedding model? 🤔 Let’s break it down! 1. What is it? Think of an embedding model as a "universal translator." It takes text, images, video, and audio data and turns them into a long string of numbers, like a unique digital fingerprint. 2. How does it work? Historically, search has been text only. Now, instead of just matching data by keyword, Gemini Embedding 2 maps multiple modalities in the same space based on meaning. It "feels" the connection between a video of a soccer goal and the words "game-winning shot" without needing tags. For example, "ocean" and "waves" are placed close together, but "ocean" and "toaster" are miles apart. 3. How can you use it? Developers have been using it to incorporate smarter search functionality into their builds. This means creating tools where you can snap a photo of a product and type "find this in yellow," or search through thousands of hours of video by describing what happens in a scene. 4. Ready to try it out for yourself? You can start using it today via the Gemini API or the Gemini Enterprise Agent Platform.

@SawyerMerritt Tesla officially ended production of the Model S and Model X in early April 2026. Only a very small number of inventory units remain worldwide. With no more new Model X coming off the line, buyers who still want one are turning to the used market.

The Tesla Model X was the fastest-selling used vehicle in the U.S. in March 2026, according to a new study by iSeeCars. On average, it found a buyer in 26 days, compared to the typical car, which usually lingers on the market for about two months. Other Tesla models are also selling faster than most used cars. The figure for the Model Y is 35 days. The figure for the Model S is 35 days.

You can now ask Gemini to create Docs, Sheets, Slides, PDFs, and more directly in your chat. No more copying, pasting, or reformatting, just prompt and download. Available globally for all @GeminiApp users.

Introducing Moonlake's 3D Agent. Our agent acts like a technical artist that can build and reconstruct articulated assets and large-scale editable scenes with hundreds of objects from a single image and can improve its generations continuously. Learn more in the thread below.

fal-roster.vercel.app , webgl experience fully made using @fal the pipeline: GPT-Image-2 (or FLUX.2 Klein 9B for the live "create your own") 🦴 meshy v6 image-to-3d takes that → rigged GLB with a baked animation 🐾 GPT-Image-2/edit on the same character → companion creature image, color-locked to the character's palette → meshy v6 again (no rig) → static creature GLB 🪨 PATINA (fal-ai/patina/material/extract) takes the character image and outputs basecolor + normal + roughness + metalness for a per-character floor that matches their environment 🎬 Seedance 2.0 fast → 8s looping background video per character 📦 everything runs through gltf-transform (resize 1024 → webp q80 → draco) for ~95% size reduction → 400-800 KB GLBs 🖥️ the front-end is a single static index.html: Three.js scene, alpha-mapped marble disc with PlaneGeometry tessellation for real displacementMap depth, mirrored reflection plane, breathing animations on companions, palette-swap via CSS custom properties on character click 🧩 the whole recipe is published as a Claude Code skill fal-regenerate-3d so you can rebuild this yourself → github.com/fal-ai-communi… ⚠ metrics shown are decorative

Someone just open-sourced a JS library that measures text without ever touching the DOM. It's called Pretext. It skips getBoundingClientRect entirely, uses canvas as ground truth, and lays out paragraphs in 0.09ms instead of triggering browser reflow. Already at 36k stars. 100% Open Source.

We're open-sourcing Hy-MT1.5-1.8B-1.25bit — a 440MB translation model that runs fully offline on your phone, supports 33 languages, and outperforms Google Translate. At 1.8B parameters, it matches commercial translation APIs and 235B-scale models on standard benchmarks. By quantizing to 1.25-bit, memory drops from 3.3GB (FP16) to 440MB — 25% smaller and ~10% faster than prior 1.67-bit approaches, with no accuracy loss. Covers 33 languages, 5 dialects, and 1,056 translation directions including minority languages like Tibetan and Mongolian. Our translation model has won 30 first-place rankings in international MT competitions and is already deployed across multiple Tencent products.🏆 📲Demo APK (Android): huggingface.co/AngelSlim/Hy-M… 🤗Hugging Face:: huggingface.co/AngelSlim/Hy-M… 🔗GitHub: github.com/tencent/AngelS… 📄Paper: arxiv.org/abs/2601.07892

In a groundbreaking medical achievement, doctors in Beijing, China, successfully replaced a large section of a patient’s spine with a custom 3D-printed titanium implant. The patient, a 41-year-old man, had been diagnosed with a severe spinal tumor that destroyed multiple vertebrae, leaving a dangerous 19-centimeter gap in his spine. Without intervention, the damage threatened paralysis and could have been fatal. Surgeons first removed the cancerous bone, then implanted a precisely engineered titanium vertebra section created using advanced 3D printing technology. What made the implant remarkable was its design. Printed from titanium powder, it perfectly matched the patient’s natural spinal curvature and featured a porous structure that allowed the patient’s own bone cells to grow into it over time. This biological integration created a stronger, more stable fusion between the artificial and natural bone. Traditional spinal reconstruction methods using rods or cages often struggle with such extensive defects. The custom 3D-printed solution offered superior anatomical fit and long-term stability. Following the complex surgery, the patient recovered well and regained the ability to walk with assistance within weeks. This pioneering operation represents a major advance in personalized medicine and demonstrates the growing potential of 3D printing to solve some of the most challenging problems in orthopedic and cancer surgery. [Hu, X., et al. (2024). 3D-Printed Patient-Customized Artificial Vertebral Body for Spinal Reconstruction after Total En Bloc Spondylectomy of Complex Multi-Level Spinal Tumors. International Journal of Bioprinting, 8(3). DOI: 10.18063/ijb.v8i3.576]

making my threejs / webgpu star renderer (planetarium?) more of a real thing. renders 44 million stars from the gaia star catalog at 120 fps. there's no LOD/culling. every star in frame is drawn. next is adding a nice API for agents to write "tours" that play automatically

Persistent memory is the Achilles heel of AI. Engramme’s Large Memory Models (LMMs) empower every app with persistent memory. Google solved search. OpenAI solved language. Engramme solved memory. Join beta: engramme.com/signup

New work with @AlecRad and @DavidDuvenaud: Have you ever dreamed of talking to someone from the past? Introducing talkie, a 13B model trained only on pre-1931 text. Vintage models should help us to understand how LMs generalize (e.g., can we teach talkie to code?). Thread:

With Voicebox, @ElevenLabs just lost its moat. → Powered by Alibaba's Qwen3-TTS for near-perfect cloning → Ships with a DAW-like "Stories Editor" → No cloud, runs locally on your machine 100% Open Source. 100% Local. Link to repo in 🧵↓

Someone just built an invisible teleprompter that hides in your MacBook notch. It's called Notchprompt. A native macOS app that floats your script right next to the camera so you can read while looking straight into the lens on Zoom calls, demos, or recordings. 100% Open Source.